Calibrated Uncertainty for Molecular Property Prediction using Ensembles of Message Passing Neural Networks

Abstract

Data-driven methods based on machine learning have the potential to accelerate analysis of atomic structures. However, machine learning models can produce overconfident predictions and it is therefore crucial to detect and handle uncertainty carefully. Here, we extend a message passing neural network designed specifically for predicting properties of molecules and materials with a calibrated probabilistic predictive distribution. The method presented in this paper differs from the previous work by considering both aleatoric and epistemic uncertainty in a unified framework, and by re-calibrating the predictive distribution on unseen data. Through computer experiments, we show that our approach results in accurate models for predicting molecular formation energies with calibrated uncertainty in and out of the training data distribution on two public molecular benchmark datasets, QM9 and PC9. The proposed method provides a general framework for training and evaluating neural network ensemble models that are able to produce accurate predictions of properties of molecules with calibrated uncertainty.

Implications of the BATTERY 2030+ AI‐Assisted Toolkit on Future Low‐TRL Battery Discoveries and Chemistries

“…The latter is critical to get sufficient data to obtain reliable statistics to derive hyperparameters and descriptors of the materials in their more complicated electrochemical environments. Developing physics-and uncertainty-aware datadriven methods [217] capable of training on such multi-sourced experimental and simulational data will strongly enhance the quality of the deep interface descriptors and features that play a critical role in shortening the path to realizing emerging battery technologies and concepts.…”